The rapid integration of artificial intelligence (AI) into educational practices has fundamentally altered how students engage with knowledge, tasks, and learning processes. As generative AI tools increasingly provide instant access to information, content generation, and problem-solving suggestions, the educational emphasis has shifted from knowledge acquisition toward the cultivation of higher-order thinking skills (HOTS), including critical thinking, creativity, metacognition, and complex problem solving. Despite this shift, the role of AI in supporting higher-order thinking remains theoretically contested.

Existing studies have produced mixed findings regarding AI’s educational value. On the one hand, AI is frequently portrayed as a powerful cognitive scaffold that expands learners’ access to information, provides multiple perspectives, and supports inquiry and reflection. On the other hand, concerns have emerged that students may rely on AI as a cognitive shortcut, outsourcing reasoning and judgment to algorithms and thereby weakening deep cognitive engagement. These contradictory outcomes suggest that the impact of AI on higher-order thinking cannot be explained by technological affordances alone.
Most prior research approaches this issue from an effect-oriented perspective, asking whether AI improves learning outcomes or specific cognitive skills. Such approaches often treat AI as an independent variable and overlook the social, organizational, and pedagogical conditions under which AI is embedded. Consequently, a key theoretical question remains insufficiently addressed: how does AI shape higher-order thinking within real educational contexts, where technology, task design, classroom structure, and student agency are deeply intertwined?

To address this gap, the present study is guided by the following research question:
How does artificial intelligence foster higher-order thinking through sociotechnical mechanisms in classroom learning contexts?

The theoretical framework of this study is grounded in sociotechnical systems theory, which emphasizes the mutual constitution of technical and social systems. From this perspective, educational technologies do not operate in isolation; their effects depend on how they are embedded within organizational structures, task designs, and human practices. Building on Leavitt’s sociotechnical model, the framework integrates four analytically distinct but interrelated components: technology, task, structure, and actors.
Within this framework, technology refers to the characteristics of AI systems, including both their generative affordances (e.g., rapid idea generation, information synthesis) and their limitations (e.g., inaccuracies, contextual misalignment). Tasks shape how AI is used by defining cognitive demands, degrees of openness, and levels of ambiguity. Classroom structures encompass organizational arrangements, instructional norms, and institutional guidance that regulate human–AI interaction. Actors, particularly students, are understood as agentic participants who adopt strategies, exercise judgment, and construct meaning in interaction with AI.The framework posits that higher-order thinking does not result from AI use per se, but emerges when these four components are productively aligned.

Overall, this study contributes a mechanism-oriented theoretical perspective on AI-supported higher-order thinking, shifting the analytical focus from technological effects to sociotechnical processes. It aims to advance conceptual clarity in the growing field of AI and education and to inform the design of learning environments that preserve and strengthen students’ cognitive agency in the age of artificial intelligence.

This study adopts a qualitative research design to capture the dynamic and context-dependent mechanisms through which AI supports higher-order thinking in classroom settings. Qualitative methods are particularly suitable for examining sociotechnical interactions, as they allow for in-depth exploration of meaning-making processes, instructional practices, and learner agency as they unfold in naturalistic environments. Research Context The study focuses on two university-level courses in which AI tools were deliberately integrated into teaching and learning activities. Rather than relying on advanced or costly intelligent systems, these courses employed widely accessible generative AI tools (e.g., large language models commonly available to students). This choice enhances the ecological validity and transferability of the findings, especially for institutions with limited technological resources. Data Collection Two primary sources of data were collected: Classroom observations: Continuous observations were conducted throughout the courses to document how AI was embedded in task design, how students interacted with AI during learning activities, and how classroom structures shaped these interactions. Detailed field notes captured instructional sequences, student–AI exchanges, peer discussions, and moments of cognitive tension or reflection. Semi-structured interviews: Interviews were conducted with both instructors and students. Interview protocols were organized around the four dimensions of the sociotechnical framework (technology, task, structure, actors), focusing on participants’ perceptions of AI use, learning strategies, decision-making processes, and experiences related to higher-order thinking. Data Analysis Data analysis followed a systematic coding process inspired by grounded theory methods, including open coding, axial coding, and selective coding. Initial codes were generated to identify recurring patterns related to AI use and cognitive engagement. These codes were then organized into higher-level categories corresponding to the sociotechnical framework. Through iterative comparison and refinement, key mechanisms explaining how AI fostered or constrained higher-order thinking were identified. To enhance trustworthiness, triangulation was employed by cross-validating findings across observation data and interview data. Data collection continued until theoretical saturation was reached, ensuring the robustness of the analytical framework.

This study is expected to yield three major contributions. First, the findings are anticipated to demonstrate that AI does not promote higher-order thinking through direct technological empowerment. Instead, higher-order thinking emerges through a sociotechnical mechanism involving the interaction of AI’s affordances and limitations with pedagogical design and student agency. In particular, AI’s imperfections—such as inaccurate outputs or contextual mismatches—are expected to function as productive triggers for critical verification and evaluative judgment when embedded in appropriately designed tasks. Second, the study is likely to show that task design and classroom structure play a critical regulatory role. Open-ended, ill-structured, and authentic tasks create conditions under which AI serves as a cognitive catalyst rather than a shortcut. Meanwhile, collaborative learning arrangements and explicit instructional guidance transform AI outputs into shared objects of reflection, comparison, and debate, thereby amplifying higher-order cognitive engagement. Third, the findings are expected to underscore the central role of student agency. Higher-order thinking develops most robustly when students adopt strategic, reflective, and iterative approaches to AI use, positioning AI as a cognitive partner rather than an epistemic authority. This shift in human–AI relations—from tool dependency to reflective collaboration—is crucial for sustaining deep learning. Overall, the study will contribute an integrated sociotechnical explanation of AI-supported higher-order thinking, advancing theoretical understanding beyond simplistic claims of technological effectiveness. Practically, it will offer design-oriented insights for educators seeking to harness AI’s potential while preserving and strengthening students’ cognitive agency in the age of artificial intelligence.

Doshi, A. R., & Hauser, O. P. (2024). Generative AI enhances individual creativity but reduces the collective diversity of novel content. Science Advances, 10(28), Article e5290. Deng, R., Jiang, M., Yu, X., Zhang, L., & Chen, Y. (2024). Does ChatGPT enhance student learning? A systematic review and meta-analysis of experimental studies. Computers & Education, Article 105224. Kong, S. C., & Yang, Y. (2024). A human-centred learning and teaching framework using generative artificial intelligence for self-regulated learning development through domain knowledge learning in K–12 settings. IEEE Transactions on Learning Technologies, 17, 1588–1599. Liu, M., Guo, S., Zhang, W., & Wang, J. (2025). How generative artificial intelligence facilitates the cultivation of learners’ higher-order thinking skills: A systematic review of 68 empirical studies. Journal of Distance Education, 43(5), 55–66. Liu, M., Zhang, L. J., & Biebricher, C. (2024). Investigating students’ cognitive processes in generative AI-assisted digital multimodal composing and traditional writing. Computers & Education, 211, Article 104977. Nguyen, A., Hong, Y., Dang, B., & Nguyen, T. (2024). Human–AI collaboration patterns in AI-assisted academic writing. Studies in Higher Education. Advance online publication. Radianti, J., Majchrzak, T. A., Fromm, J., & Wohlgenannt, I. (2020). A systematic review of immersive virtual reality applications for higher education: Design elements, lessons learned, and research agenda. Computers & Education, 147, Article 103778. Rafner, J., Beaty, R. E., Kaufman, J. C., & Silvia, P. J. (2023). Creativity in the age of generative AI. Nature Human Behaviour, 7(11), 1836–1838. Saritepeci, M., & Yildiz Durak, H. (2024). Effectiveness of artificial intelligence integration in design-based learning on design thinking mindset, creative and reflective thinking skills. Education and Information Technologies. Advance online publication. Sharma, D. (2024). Critical thinking and problem-solving in the age of ChatGPT: An experiential blogging project. Business and Professional Communication Quarterly, 87(4), 630–653. Wu, T. T., Lee, H. Y., Li, P. H., Chen, S. C., & Heh, J. S. (2024). Promoting self-regulation progress and knowledge construction in blended learning via ChatGPT-based learning aid. Journal of Educational Computing Research, 61(8), 3–31.